[maerF0x0]

I've noticed that chatgpt doesnt seem to be very good at understanding elapsed time. I have some long running threads and unless i prompt it with elapsed time ("it's now 7 days later") the responses act like it was 1 second after the last message.

I think this might be a good leap for agents, the ability to not just review a doc in it's current state, but to keep in context/understanding the full evolution of a document.

[wat10000]

They have no ability to even perceive time, unless the system gives them timestamps for the current interaction and past interactions.

[multiplegeorges]

Which seems like a trivial addition if it's not there?

[wat10000]

It is, but now you're burning a bit of context on something that might not be necessary, and potentially having the agent focus on time when it's not relevant. Not necessarily a bad idea, but as always, tradeoffs.

[HankStallone]

I've noticed the same thing with Grok. One time it predicted a X% chance that something would happen by July 31. On August 1, it was still predicting the thing would happen by July 31, just with lower (but non-zero) odds. Their grasp on time is tenuous at best.